molten salt leblanc

8/12/2019 Molten Salt LeBlanc

1/14


2/14

Author's personal copy

Nuclear Engineering and Design 240 (2010) 16441656

Contents lists available atScienceDirect

Nuclear Engineering and Design

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / n u c e n g d e s

Molten salt reactors: A new beginning for an old idea

David LeBlanc a,b,

a Physics Department, Carleton University, 1125 Colonel By Drive, Ottawa, ON, Canada K15 5B6b Ottawa Valley Research Associates Ltd., Canada

a r t i c l e i n f o

Article history:Received 24 September 2009

Received in revised form 3 December 2009

Accepted 18 December 2009

a b s t r a c t

Molten salt reactors have seen a marked resurgence of interest over the past decade, highlighted bytheir inclusion as one of six Generation IV reactor types. The most active development period however

was between the mid 1950s and early 1970s at Oak Ridge National Laboratories (ORNL) and any new

re-examination of this concept must bear in mind the far different priorities then in place. High breeding

ratios andshort doubling times were paramountand this guidedthe evolution of theMoltenSalt Breeder

Reactor (MSBR) program. As the inherent advantages of the molten salt concept have become apparent

to an increasing number of researchers worldwide it is important to not simply look to continue where

ORNL left off but to return to basics in order to offer the best design using updated goals and abilities.

A major potential change to thetraditional SingleFluid, MSBR design and a subject of thispresentation

is a return to the mode of operation that ORNL proposed for the majority of its MSR program. That

being the Two Fluid design in which separate salts are used for fissile 233 UF4 and fertile ThF4. Oak Ridge

abandoned this promising route due to what was known as the plumbing problem. It will be shown

that a simple yet crucial modification to core geometry can solve this problem and enable the many

advantages of the Two Fluid design. In addition, another very promising route laid out by ORNL was

simplified SingleFluid converter reactors that could obtain far superior lifetime uranium utilization than

LWR or CANDU without the need for any fuel processing beyond simple chemistry control. Updates and

potential improvements to this very attractive concept will also be explored.

2010 Elsevier B.V. All rights reserved.

1. Introduction

A molten salt reactor (MSR) is one in which fluorides of fissile

and/or fertile elements such as UF4, PuF3 and/or ThF4 are com-

bined withcarrier saltsto formfluids. Single Fluiddesigns have both

fertile and fissile combined in one salt, whereas the lesser known

Two Fluid design has separate salts for fissile (233UF4) and fertile

(ThF4). Typical operation sees salt flowing between a critical core

and an external intermediate heat exchanger. A secondary coolant

salt then transfersheat to steam or closedgas cycle.The vast major-ity of work has involved fluoride salts as corrosion resistant alloys

have been shown to be compatible with these salts. Chloride based

salts have also been proposed, especially for fast breeder designs,

buthave unique problems andlack operational experience to draw

upon. Designs specifically for the Th-233U cycle using fluoride salts

have recentlybeen termed Liquid Fluoride Thorium Reactors (LFTR)

and is a term becoming more widely used.

Corresponddence address: Physics Department, Carleton University, 1125

Colonel By Drive, Ottawa, ON, Canada K15 5B6. Tel.: +1 613 520 2600;

fax: +1 613 746 9632.

E-mail address:[email protected].

The most common carrier salt proposed are mixtures of

enriched (>99.99%) 7LiF and BeF2 termed flibe. Mixtures with

upwards of 14% ThF4 and/or UF4have melting points below a tra-

ditional limit of 525 C which gives adequate margin for use with

the high nickel alloy Hastelloy N. This alloy is rated for use up to

704 C with these molten salts and options exist to employ newer,

ASME Section III qualified materials such as Alloy 800H clad with

Hastelloy N. It has been suggested that carbon based structures

or refractory metals could be used throughout the primary loop

including heat exchangers which would allowhigher peak temper-atures and a wider set of carrier salts with higher melting points.

However, in the following presentation the more conventional

temperature limits will be assumed.

Molten salt reactors have numerous operational and safety

advantages over solid fuel designs. A detailed review is beyond the

scope of this presentation but briefly:

- Fluid nature of the fuel means meltdown is an irrelevant term

and allows the fuel salt to be automatically drained to passively

cooled, critically safe dump tanks.

- Fission products eitherquickly form stablefluoridesthat will stay

withinthe salt during anyleak or accident or are volatile or insol-

uble and are continuously removed. Noble gases bubble out and

0029-5493/$ see front matter 2010 Elsevier B.V. All rights reserved.

doi:10.1016/j.nucengdes.2009.12.033


3/14


D. LeBlanc / Nuclear Engineering and Design240 (2010) 16441656 1645

are stored outside the reactor loop. Noble and semi noble met-

als will plate out on metal surfaces and/or can be collected by

replaceable high surface area metal sponges within the loop.

- The continuous removal of the noble gas Xenon means that there

is no dead time of the reactor after shutdown or a power

decrease that most solid fueled reactors must deal with due to

the production of135

Xe from the decay of135

I. As well, no excessreactivity need be in place to deal with such events.

- Most MSR designs have very strong negative temperature and

void coefficients which act instantly, aiding safety and allowing

automatic load following operation.

- Only a low pressure vessel is needed as the salts run near atmo-

spheric pressure as opposed to the thick walled vessels needed

for LWR or PBMR. No water or sodium means no possible steam

explosion or hydrogen production within the containment. In

designs without graphite moderator, there is not even com-

bustible material present.

- Molten fluoride salts in general are excellent coolants, with a

25% higher volumetric heat capacity than pressurized water and

nearly5 times that of liquidsodium. This results in more compact

primary loop components like pumps and heat exchangers. Theyalso have a much lower thermal conductivity than sodium which

avoids thermal shock issues.

- Fissile concentrations are easily adjusted on a continuous basis

meaning no excess reactivity and no need for burnable poisons.

A single control rod for shutdown and start-up is often included

but is not strictly necessary given the ability to drain fuel out of

the core into critically safe storage tanks.

- Utilization of the thorium to 233 U cycle produces several orders

of magnitude less transuranic wastes than a conventional once

through cycle and significantly less than even a U-Pu fast breeder

(based on 0.1% losses during fuel processing). This leads to

radiotoxicity of waste being less than equivalent uranium ore

levels within a few hundred years.

- Fuel processing and utilization of thorium permits break even

breeding with ease and ability to reach a breeding ratio of 1.06

or even up to 1.13. Adding238U to denature the uranium content

and still break even is also possible.

- Break even operation only requires approximately 800kg of

thorium per GW(e) year added simply as ThF4. Start-up fis-

sile requirements can be as low as 200kg/GW(e) or upwards

of 5.5tonnes in harder spectrum designs, with 7001500kg

more common. Thorium start-up inventory varies from 50 to

200 tonnes per GW(e).

- Terrestrial thorium is 3 times as abundant as uranium with large

proven reserves even with the small current usage. As exam-

ple, a single new deposit in Lemhi Pass Idaho has recently added

600,000tonnesto theworldsproven reserves increasing reserves

by 50%.

- Without any fuel processing, MSRs can run as simple converterswith excellent uranium utilization even on a once through cycle

(less than 1/3 that required for LWR).

- MSRs offer large advantages for the destruction of transuranic

wastes from traditional once through reactors. TRUs may also be

used as start-up fissile inventory in most designs.

2. Background and history

Molten salt reactors were developed primarily at Oak Ridge

National Laboratories (ORNL) beginning in the late 1940s. Almost

30 years of funded research and development followed a design

evolution leading to the adoption in the late 1960s of what is

known as the Single Fluid, graphite moderated MSBR (Weinberget al., 1970). What is important to realize is that this evolution

was guided by priorities and limitations that are far different than

now exist. In particular, the overwhelming priority given to the

MSBR program was a minimization of the doubling time, the time

to breed the start-up fissile inventory of the next reactor. The two

routes for this are decreasing the necessary start-up inventory and

increasing the breeding ratio. This mandated priority was due to

the early belief that nuclear power would follow an exponential

growth and that uranium supplies were severely limited. Associ-ated to this was that the MSBRs main competition was the well

funded liquid sodium cooled fast breeder whose potential doubling

time was impressive at the time but later would be greatly length-

ened due to the need to soften the neutron spectrum for safety

reasons.

The development of MSRs can be categorized into four eras.

The very first work was in support of the Aircraft Reactor Pro-

gram for the U.S. Air Force. The concept being an onboard MSR to

replace combustion heat for the jet engines of bombers. While this

project did not lead to an operational bomber, it did lead to a large

knowledge base being developed and to a successful test reactor,

the Aircraft Reactor Experiment. The ARE was a high temperature

reactor with a peak temperature of 860 C employing a NaF-ZrF4

carrier salt and fuelled with highly enriched

235

U. Clad blocks ofBeO provided moderation.

A true molten salt program began in the mid 1950s to investi-

gate power station designs to be either simple converter reactors

or potential break even or breeder reactors on the Th-233U cycle.

Thesetook theform of verysimplesphere within sphere, tworegion

designs as depicted inFig. 1. A central molten salt core would con-

tain fissileand also usually fertile material andwas surrounded by a

ThF4containing blanket salt. The7Li-BeF2carrier salt itself is capa-

ble of modestly efficient moderation of neutrons such that a wide

variety of spectrumsfromsoft epithermal to fairlyhardwereexam-

ined. A HastelloyN barrier between regions was to be employed. In

later years it would be found that such a usage within the core and

thus within a high neutron fluence would have a limited lifetime

dueto heliumembrittlement from (n,alpha)reactions in nickeland

boron contaminants.

Fig.1. Depicts OakRidgestwo regionconceptof the1950s.ReproducedfromORNL

2474.


4/14


1646 D. LeBlanc / Nuclear Engineering and Design240 (2010) 16441656

By 1960 it was determined that graphite was indeed capable of

long term interaction with the salts and appeared to offer a way

of more efficient neutron moderation and the potential to lower

fissile starting loads. The use of graphite within the core also gave

researchers at ORNL the hope that a true Two Fluid design could be

made practical.

A Two Fluid design uses a fuel salt that carries the fissile233

UF4and a separate blanket salt for the fertile ThF4 . As

233 U is produced

in the blanket, it is transferred to the fuel salt by a simple fluorina-

tion process. Fluorination is accomplished by bubbling fluorine gas

through the salt which converts UF4to volatile UF6. Gaseous UF6is

collected and converted back to UF4 by a well established process

before being added to fuel salt.

The main advantage of a Two Fluid system is that processing

out fission products from the fuel salt is greatly simplified by the

absence of thorium. The prime method being known as vacuum

distillation (Scott et al., 1966) developed in 1964. After removal

of all UF4 , the carrier salt would be evaporated off and recycled,

leaving most fission products behind.

Furthermore, the neutron losses to 233Pa can be minimized as it

is effectively diluted in the much larger volume of blanket salt andexperiences a lower neutron flux. 233Pa is the 27 day half live inter-

mediate which decays to 233U followingits production bya neutron

absorptionin 232Th.ThusTwoFluiddesignsdonotrequireanycom-

plex and rapid processing to remove Pa to allow it to decay to 233U

as is oftenthought necessary. ORNL often didproposePa removal in

much of their Two Fluid work but only to slightly improve breed-

ing ratios (1.07 vs. 1.05). There would be little modern incentive

to include Pa removal and large economic and other benefits from

avoiding this process.

In this new concept, the graphite was to be used to internally

separate the fuel salt and blanket salt withinthe core (Robertson et

al., 1970) as shown inFig. 2. However this use of graphite plumb-

ing of hundreds of tubes proved a large challenge as a crack in

any tube might mean replacement of the entire core and vessel

dueto itscomplexity. Underfast neutron fluence, graphite will first

contract and then expandwhich leads to a limited lifetime depend-

ing on power density and graphite temperature. Fairly high power

density cores were preferred by ORNL giving typical graphite life-

times of 4 years before reaching a peak fluence of 3 1022 n/cm2

(E> 50 keV). A further complication is the contraction and expan-

sion of graphite also led to a troublesome changing ratio of fuel

to blanket salts within the core which could adversely affect reac-

tivity. It is a testament to the advantages of Two Fluid design that

ORNL stuck with this complex design until the late 1960s despite

these often termed plumbing problems.

Also during the 1960s, the highly successful test reactor, the

Molten Salt Reactor Experiment was constructed and operated. It

was an 8 MW(th) design chosen to be a Single Fluid for simplicity.

It operated for nearly 5 years with great success. Two unknownissues with Hastelloy N did surface, one was corrosion induced

by the fission product tellurium and the other was irradiation

damage caused by (n,alpha) reactions in nickel and boron con-

taminants. In subsequent years of the program, these issues were

largely addressed by modifying the alloy makeup of the Hastelloy

and the reduction potential of the salt accomplished simply by the

occasional additions of metallic beryllium. It is likely though that

Hastelloy N has a limited lifetime if used within the full neutron

flux of the core. Use in the outer vessel walls and heat exchangers

should pose little problem but substantial work will be required in

order to qualify any new alloys for ASME Section III use.

In 1968 the development of a new fission product removal

methodthatcouldfunctionin the presence of thorium andan inge-

nious technique to limit neutron leakage in a Single Fluid designalong with the plumbing problems of the Two Fluid design men-

tioned above led ORNL to abandon its Two Fluid breeder concept.

Fig. 2. Generalized depiction of the 1960sintermixed Two FluidMSBR design using

internal graphite plumbing. Reproduced from ORNL 4528.

Liquid Bismuth Reductive Extraction (McNeese, 1971), while not a

simple technique, was shown to be able to differentiate between

thorium and the very important rare earth fission products. Single

Fluid designs with both thorium and 233U within the same salt also

mean that 233Pa will be a larger drain on the neutron economy.

In ORNLs MSBR design the Breeding Ratio would drop from 1.06

down to 1.01 if233 Pa was not removed on a 310 day time frame

(Rosenthalet al.,1970). Given thestrong emphasis placed on breed-

ing at the time they felt Pa needed to be removed and allowed to

decay outside the core before being returned as 23 3U. Pa removal

is also accomplished by Liquid Bismuth Reductive Extraction. It is

worth noting that while this technique for removing fission prod-

uctscan functionwith thoriumpresentin thesalt, it works farbetterwithout it as thorium is chemically similar to the rare earth fission

products. Thus there is still a great advantage to Two Fluid designs

when it comes to fission product removal.

Another area of advance that made a Single Fluid graphite mod-

erated concept appear more attractive was the idea of employing

an under moderated outer zone of the graphite core in which a

higher ratio of fuel salt to graphite hardened the spectrum and led

to the region being a net absorber of neutrons. This results in a

greatreduction in neutron leakage and/or reflector losses. Typically

about 12% salt volume proved best in the inner zone while 37% salt

volume was proposed for the outer absorbing zone. The 37% value

waschosen to allow thepossible use of graphite pebbles. A random

bed of pebbles occupying 63%of the volume (although typically 61%

is now generally assumed).The end result was what might be termed the 4th era of

molten salt development and produced the traditional Molten


5/14



Fig. 3. The 1970s Single Fluid, graphite moderated Molten Salt Breeder Reactor. Breeding Ratio of 1.06, with a specific inventory of 1500 kg/GW(e). Reproduced from ORNL

4812. Freeze plug and drainage system to dump tanks not shown.

Salt Breeder Reactor design (Weinberg et al., 1970),Fig. 3, which

has been included as one of six Generation IV reactors rela-

tively unchanged. This design has a starting fissile load of 1500 kg233U/GW(e), a Breeding Ratio of 1.06, a 20 year doubling time, a

20 day cycle for fission product and 3 day cycle for 233 Pa removal

(both by Liquid Bismuth Reductive Extraction). Sealed graphite is

used to limit Xenon retention with a 4 year cycle between graphite

replacement. The 1000MWe design has an average power density

of 22 MWth/m3 and a core size of 5 m diameter and 4.3 m height.

In the early 1970s however, for reasons many would argue farmore political than technical, the MSBR program was terminated

by the AEC. Corrosion issues that were already well on their way to

being addressed were the main faults mentioned (Shaw, 1972) and

which unfortunately have become engrained in memory. Certainly

significant work is still needed to qualify materials for the primary

loop including heat exchangers but the materials challenge is likely

far less than the AEC report implied. The sudden termination of

funding also meant the Single Fluid design became the textbook

design and little mention of alternatives has appeared until very

recent years. Oak Ridge did continue a modest program until the

early 1980s, with a greatly increased value placed on maximizing

proliferation resistance. ORNL examined operating on denatured

cycles in which all uranium stayed below the weighted average of

12% 233U and/or 20% 2 35U. The results were surprisingly success-ful, with two routes examined, both termed DMSR for Denatured

Molten Salt Reactor. The first a DMSR break even design (Engel

et al., 1978) with similar fission product processing to the MSBR

and showed that break even breeding could be accomplished even

while remaining in a denatured state using depleted 23 8U and Th

as the fertile makeup. The second was a greatly simplified DMSR

converter design called the 30 Year Once Through Design ( Engel

et al., 1980) without any fuel processing beyond chemistry control

for a full 30 years while still maintaining a very high conversion

ratio and excellent uranium resource utilization. Both designs also

featured larger, low power density cores that gave a full 30 year

lifetime of the graphite.Over the next 20 years however, little advance was made as

funding worldwide was virtually non existent. Many vocal propo-

nents worldwide such as Charles Forsberg and Uri Gat of ORNL and

Kazuo Furukawa in Japan at the very least kept interest active dur-

ing lean times for all nuclear designs (Forsberg, 2006b; Furukawa

et al., 1990; Furukawa et al., 2008; Gat et al., 1992; Gat et al., 1997).

3. A resurgence of interest

Theselection of moltensalt reactorsas one of thesix Generation

IV reactorsin 2002 hascertainly contributed to an increase in inter-

est.Much recent activityhas alsobeen basedon molten saltreactors

acting as transuranic waste burners. There are numerous advan-

tages in this regard as no solid fuel elements require fabrication,all minor actinides can be included and there are no concerns over

varyingisotopiccompositions. MostinitialTRU burner worklooked


6/14


7/14



components willhave a greatdeal of similarity. Costand complexity

of the core itself along with the costs and R&D needs of any on site

or centralized fuel processing are likely the major differentiators.

It is highly likely that different MSR designs may prove opti-

mal for different nations, vendors or utilities. Thus it is likely a best

optionto thoroughly investigatemore than onedesign. Thefollow-

ing representsmajor new areasbeing proposed and investigated bythe author.

5. Solving the two fluid plumbing problem

All original fluid fuel reactor designs involved utilizing two

zones, a central core or seed zone surrounded by a fertile blan-

ket (i.e. thorium). For the 2 33U-Th cycle, the core might contain a

mix of233 U and Thorium or in some cases only233 U. ORNL differ-

entiated between these twocasesby thetermsTwo Fluidfor only

fissile233 U in the core and 1 and 1/2 Fluid designs if the core salt

contained thorium as well.

Early in development, the advantages of a Two Fluid design

became evident. If the core salt lacked thorium, it would be far

easier to process out fission products. However a core with onlyfissile salt will have a quite small critical diameter if the fissile con-

centration is also kept high enough to limit neutron losses to the

carrier salt and/or graphite. This critical diameter is on the order of

1 m foreither pure salt or heterogeneous cores employinggraphite.

Since power density has limitations for many reasons, such a small

core would have very limited power output. ORNLs solution was

to use plumbing to intermix the Two Fluids within the core zone.

However, as previously reviewed, this proved unmanageable.

A new solution proposed by the author may in fact be sur-

prisingly simple (LeBlanc, 2008). Traditionally reactor cores are

spherical or short cylinder primarily to minimize neutron leak-

age. If a Two Fluid design has an encompassing blanket, leakage is

not an issue. The simple solution thus proposed is a core geometry

switchto increase the power producing volume while maintainingthe required small critical diameter.

As a first approximation the critical diameter will be the ratio

of the Buckling constants between the given geometries. Thus, for

the same fuel salt and/or graphite combination,a long cylinder will

have a critical diameter approximately 77% that of a sphere. If a

specific combination of fissile concentration, graphite and carrier

salt gives a critical diameter of 1 m fora sphere, then a long cylinder

would have critical diameter of 0.77 m.

The great advantage of going to an elongated cylinder or slab

is the fact that a practical total power can now be obtained with-

out intermixing but by simply extending the length of the core. A

single barrier needs to be maintained between the core and blan-

ket regions, which is far less complex than the internal plumbing

of fuel and blanket salts in ORNLs Two Fluid designs. In terms of

end plenums on these cylindrical cores, a possible arrangement

would be to taper the ends to a sub critical diameter while still

surrounded by the blanket salt, see Fig. 4. This will all buteliminate

Fig.4. Generalized depictionof an elongatedcylindricalTwo Fluidcorewith encom-

passing blanket salt. Inlet/outlet for blanket salt cooling are not shown.

leakage of neutrons. While modeling efforts are ongoing, previous

calculations from ORNL work of homogeneous designs of the late

1950s and Two Fluid graphite work of the 1960s can be used to a

significant degree to predict characteristics.

Such a design will have a strong negative temperature and void

coefficient for the fuel salt which is true for any Two Fluid design.

A major improvement over ORNLs internally separated Two Fluiddesign is that the blanket should also have negative coefficients.

This is due to the fact that the outer blanket acts as a weak neutron

reflector. Thus lowering its density decreases this reflective quality

and lowers reactivity in the core. It should be mentioned that to

attain this desired negative temp/void blanket salt coefficient, no

graphite or other reflective materials should be used in the blanket

zone. This is easily accomplished.

As with any fluid fuelled, two zone design, the leakage of

core fluid into the blanket must be guarded against. The simplest

method, proposed forall ORNL TwoFluiddesigns is to runthe blan-

ketfluidataslightlyhigherpressure.Astheblanketsaltisfardenser

than the core salt, hydrostatic pressure alone accomplishes this.

Thus any leak through the barrier will add fertile to the core and

lowers reactivity.There are advantages to employinggraphite moderation includ-

ing very low fissile specific inventories and providing a built in

structure to aid in the barrier between core and blanket. However,

the much lower overall power density of graphite designs results

in the need for greater overall core volume to attain power plant

levels. This likelymeans multiple units perplant butthisalso brings

other operation advantages. The limited lifetime of graphite due to

fast neutron damage would also entail periodic replacement as is

true of most MSBR designs. Thesmalldimension and multiple units

should assist in this operation.

The total volume of salt and the fissile molar concentration

dictate the specific inventory. For a graphite moderated design it

should be possible to reach 0.15% 233UF4 molar concentration or

even 0.1% or lower and still break even (i.e. breed just enough 233U

to match consumption). Taking into account salt volume needed

out of core leads to a conservative estimate of 20 m3 per GW(e) as

adopted in recentFrench studies andearlyORNLwork with a lower

limit of perhaps 10 m3 given the possible use of new compact heat

exchangers to greatly reduce out of core volumes. These estimates

give a potential lower limit of start up fissile inventory of a mere

150 kg/GW(e) with 400 kg/GW(e) being a more conservative goal.

For comparison ORNL Two Fluid work was about 700kg/GW(e),

ORNL SingleFluid1500kg/GW(e), anLWR is 35tonnes/GW(e) and

liquid metal cooled fast breeders 1020 tonnes/GW(e).

More impressive perhaps are the possibilities withhomogenous

designs lacking graphite moderator. With the entire volume of the

core now beingpowerproducing salt, theneeded volumeis farless.

Single cores for 1000 MW(e) are readily attainable although there

are still advantages to smaller unit sizes. Without graphite moder-ation, the assumption is often made that this means a much higher

specific inventory and a quite hard spectrum. However, the carrier

salt itself is a fairly effective moderator and a wide variety of fissile

concentration and neutron spectrums are in fact attainable. Recent

French work requires a high specific inventoryof 5.5tonnes/GW(e)

partly due to only having a radial blanket in the TMSR design.

Attempting a much lower concentration would see a significant

increase in neutron losses to the top and bottom reflectors due to

the longer migration paths of neutrons before absorption. With a

fully encompassing blanket, this is not an issue.

ORNL calculations (Alexander et al., 1959) from the spherical

cores of the 1950s design provide an excellent tool for estima-

tion, seeTable 1. While the accuracy of such early data must of

course remain suspect, it is hoped adequate for at least cursoryinvestigations. This study assumed a 8.5 mm (1/3 in.) thick Hastel-

loy N barrier for cores up to 3.7m (12 feet) diameter, thus for


8/14



Table 1

Initial state nuclear characteristics of spherical two region, homogeneous, molten fluoride salt reactor with 233 U from ORNL 2751 (1959). Values in italics are projected by

the author.

Inner core diameter 91 cm 122 cm 122 cm 183 cm 244 cm

Thorium in fuel salt 0% 0% 0.25% 0% 7%233 UF4in fuel salt 0.592% 0.158% 0.233% 0.048% 0.603%

Neutrons per absorption in 233 U Be, Li and F in Fuel Salt 0.0639 0.1051 0.0860 0.318 0.078

Hastelloy N Core Wall (8.5 mm) 0.0902 0.1401 0.1093 0.1983 0.025Li and F in Blanket salt 0.0233 0.0234 0.0203 0.0215 0.009

Leakage 0.0477 0.0310 0.0306 0.016 0.009

Neutron yield 2.1973 2.1853 2.1750 2.2124 2.200

Median fission energy 174 ev 14.2 ev 19.1 ev 0.33 ev 243 ev

Initial breeding ratio 0.9722 0.8856 0.9288 0.6586 1.078

Projected B.R.thinner walla 1.060 0.9836 1.011 0.7722 1.099

Projected B.R.carbon wallb 1.105 1.054 1.066 0.8714 1.112

a Projected assuming a thinner Hastelloy core wall of 1/6 inch (4.2 mm) and 90% leakage reduction by using a thicker blanket.b Projected assuming a graphite or carboncarbon core wall and 90% leakage reduction by using a thicker blanket salt.

much smaller cylindersa thinner wall shouldsuffice. Thestudyalso

assumed a 60 cm (2 foot)blanketwhichallowedsignificantneutron

leakage in some cases, expanding this to 100 cm should convert a

high percentage of those losses into thorium absorptions.

The values ofTable 1give the initial breeding ratios, thus no

losses to fission products or protactinium. ORNL also projected

(MacPherson et al., 1958) long term breeding ratios for the 8 foot

core case in detail. Even with a relatively long 1 year processing

time for fission product removal and no protactinium separation,

the breeding ratio only dropped from 1.078 to 1.044.

Taking the 3 foot (91cm) case as example, this would equate

to a 70 cm wide cylindrical core. A modest average power density

of 200MW/m3 still gives impressive results. Using the standard

ORNL 140K inlet/outlettemperature changeand a salt speedin core

of 2 m/s gives a 505 MW(th) output from a 6.6m long, 0.7 m wide

core. At 44.4% fora steam cycle,this is 224MW(e) with a somewhat

higher output if a gas cycle is used. Including a meter thick blanket

and outer vessel wall still results in a simple to manufacture design

that can fit within a tractor trailer for transport.It mustbe noted thatHastelloy N atthe time of these early stud-

ies was thought to be good for 10 to 20 years in core. Neutron

induced helium embrittlement discovered in the MSRE means that

HastelloyN mightnot have a long lifetime in the full fluxof the core.

ORNL had success in limiting this damage by modifying the alloy

makeupandthistrendhasbeencontinuedin recentFrenchworkon

the TMSR where they are now proposing a new Hastelloy alloy for

both their radial blanket zone and axial reflectors (Merle-Lucotte,

2008a; Cury, 2007). Potentially a much superior metal barrier is a

high molybdenum alloy which is known to have a much greater

tolerance to neutron damage (Zinkle and Ghoniem, 2000). It has

been suggested for use not only in molten salt fission designs but

also for the barrier between plasma and a LiF-BeF2 coolant salt in

fusion studies. Furthermore, less expensive iron alloys includingthe common stainless steels 304and 316have also shown promise

both for corrosion and withstanding neutron irradiation if peak

temperatures are lowered somewhat (Zinkle and Ghoniem, 2000;

Koger, 1972; Keiser et al., 1977; DiStefano et al., 1995) , perhaps

to 550 or 600 C. Given the simplicity of the core wall and outer

vessel combination it is also not unreasonable to assume that peri-

odic replacement even as short as annually could be still practical

and economic.Fig. 5 shows and embodiment to potentially ease

periodic replacement.

Carbon based material or a simple graphite tube might also

prove a workable barrier. The limited lifetime of graphite is well

documented and would likely require periodic replacement. Sili-

con impregnated carbon-carbon composite is a leading candidate

in fusion studies with molten salt coolants but questions remainregarding compatibilitywith the salts. There are thus manychoices

for a barrier material but it should be emphasized that this issue is

of central importance to this proposed design as well as any other

MSR design looking to include a thorium salt blanket region.

This basic design premise has of course numerous other possi-

ble embodiments. A major area is adding either thorium or 238U

to the fuel salt, both of which would increase the critical diameter

andlower fluxlevels atthe barrier (i.e. shorter, wider cores). Adding

thorium would appearto go against the fuel processing advantages

of a Two Fluid design but given the low cost and abundance of tho-

rium it maybe optimal to simply let modest amounts be processed

out with the fission products. Adding 238U to both core salt and

Fig. 5. An embodiment of a Tube within Shell core design employing a calandria

arrangement to accommodate thermal expansion and/or tube replacement.


9/14



blanket is of course a way to operate in a fully denatured state. For

the Two Fluid designs proposed, each time the blanket salt is flu-

orinated to remove the produced 23 3U, an appropriate amount of238UF4 is added back to the thorium blanket. The amount added

is enough to assure that all removals of UF4 from the blanket will

have the uranium in a denatured state.

6. Start-up on low enriched uranium

If a truly large scale deployment of molten salt reactors is to

take place, start-up fissile material may become an issue. While

the starting loads are typically far less than liquid metal cooled

fast breeders or even LWRs, no significant quantities of2 33U now

exist. While spent fuel transuranics are an obvious fissile source,

even these may be limited and they also pose a possible prolif-

eration and public acceptance issue of shipping TRUs to start up

reactors. Highly enriched 235U would function and early work at

ORNL typically assumed its use. However large scale production

and shipments of such HEU would be ruled out today. Start-up of

denaturedconverter designs or denaturedbreak evendesigns obvi-ously poses noproblem inusingLEU (


10/14



cle, they do not improve dramatically. Thus the great advantage

of molten salt converter reactors is that all plutonium produced

stays within the core for the full lifetime where it is mostly con-

sumed.

In terms of long term radiotoxicity of wastes, these converter

designs also perform remarkably well. All transuranics remain in

the salt during operation and will not reach high concentrationsdue to the very large cross sections for fission and/or absorption.

At the end of 30 years there is only about 1000kg present. It is

presumed prudent to perform a one time only process to remove

these transuranics for recycle into the next core salt. If this is done

and a typical processing loss of 0.1% is assumed, this represents

a mere 1 kg of TRUs going to waste over 30 years which is about

a 10,000 fold improvement over LWR once through. An average

then of about 30g per GW(e)-year of TRU waste is even better than

most pure Th-233U MSR concepts that process smaller amounts of

TRUs but do so much more frequently. It should be mentioned that

removing and reusing TRUs does not involve isolating plutonium.

The likely process would be Liquid Bismuth Reductive Extraction

and Pu would remain with Am, Cm, Cf and the fission product

zirconium.Such a recycling of TRUs also has a positive impact upon life-

time uranium utilization. In ORNLs studies, at the end of the 30

year cycle, there would be 736 kg of Pu of which 331kg would be

fissile 239Pu and 2 41Pu. Along with this would be 136 kg of 237Np

and small amounts of Am and Cm (quantities not reported). There

wouldalso be enoughuraniumleft tostartthe next cycle with clean

salt, leaving the TRUs as source of fissile top up. This should save

approximately 510% on the lifetime uranium utilization. Thus the

DMSR without any recycling is about 1800tonnes orelifetime, with

U only recycling is 1000 tonnes andwith U + TRU recycling down to

perhaps 900950 tonnes ore.

Another aspect of this design which is perhaps little known

is that operation as a converter with 238U present leads to much

improvedreactivity coefficients. As mentioned, recent French work

(Nuttin, 2002) brought to light that the overall global tempera-

ture reactivity coefficient of the ORNL MSBR design was actually

a slightly positive +0.6pcm/K as opposed to the small negative

value of2.4pcm/K that ORNL calculated (pcm = 105 k/k). This

appears to have been due to early calculations not properly tak-

ing heterogeneous effects into account. For the converter design

though, ORNL calculated (Engel et al., 1980) a muchhigher value of7.2pcm/K such that it should remain strongly negative even with

up to date corrections.

Starting with ORNLs already attractive DMSR converter reactor

design, what are some possible improvements? Before a review of

proposed modifications a few last points should be mentioned.

All SingleFluid designs that do notattempt to remove 233Pa will

experience some neutron losses to Pa absorptions. Removing the

need to break even on breeding means this is not of great concernbut there are still two routes to lower these losses with the goal

of improving uranium utilization. The first being a larger specific

fissile inventory to out compete for neutrons. The larger cores and

salt volumes of ORNLs converter design already accomplish this

(i.e. 3450kg fissile versus 1500 kg fissile for the MSBR). However

the neutron spectrum also has a major effect. Lowest losses are

seen with either a well thermalized spectrum or alternatively with

a muchharder spectrum. ORNLs DMSR converter design employed

a softer spectrum than the MSBR for this reason.

The other extreme to lower Pa losses is attempting to keep the

spectrum quite fast. There are other benefits to a faster spectrum

including an improved capture to fission ratio for 239Pu and fast

fission bonus from 238U. It should be noted though that due to

very strong inelastic scattering cross sections for fluorine, suchbenefits are limited as these interactions quickly bring neutrons

well below the MeV range. For example, even the TMSR design of

France with a very high fissile concentration and no graphite sees

a 239 Pu capture/fission ratio of 0.53 (Merle-Lucotte, 2008b) which

is only modestly better than that of well thermalized designs such

as the MSBR and much higher than the value of 0.25 which can

be obtained with similar fissile concentrations to the TMSR but

with chloride based carrier salts. This latter value is taken from

the work on the REBUS 3700 chloride salt design (Mourogov andBokov, 2005).

Costs of the carrier salts are also an issue since typically

these converter designs will employ more. For 1 GW(e), the DMSR

requires 100 m3,theMSBR54m3 andtheTMSR20m3.Thestandard

carrier salt flibe requires enriched lithium which unfortunately is

hard to judge costs and production potential for large amounts.

100m3 of flibe will contain about 30 tonnes of 99.995% 7Li with

previous cost estimates being from 120 to 800 $/kg. Even several

thousand dollars per kg would still not prohibit its use but it is

certainly an important issue. Beryllium metal is also an expensive

item but it is a fortunate coincidence that an intermediate stage

of all beryllium metal production is BeF2 which means this form

will have a much lower cost and have ease of large scale produc-

tion.The total volume of carrier salt is also of importance in deal-

ing with loss of cooling or transient power events. A large volume

gives the benefit of a massive heat sink, especially in combina-

tion with the heat sink of the even more voluminous graphite.

While freeze plugsand automaticdraining to tanks designedto deal

with decay heat provide the main line of defense, it is reassuring

that decay heat would take hours to bring the system anywhere

near the boiling point of the salt even if zero heat removal is

assumed.

7.2. Simple changes to the DMSR converter

Oak Ridges work on the DMSR converter was not subject to a

great level of optimization of fundamental parameters. The sim-

plest possible change would be to vary the fissile plus fertile

concentration in the salt. The DMSR converter had 9.5% molar con-

centration of (U+ Th)F4 in the flibe salt. This concentration can

easily be lowered or raised to upwards of 1418% without rais-

ing the melting point beyond traditional limits for Hastelloy N

use. Lowering the concentration would mean a smaller starting

fissile load, important for initial costs but this would lower the

long term conversion ratio and mean a higher lifetime uranium

requirement. Going to a higher fissile specific inventory, say to the

same 5 tonnes 235U perGW(e)LWRs currently require would mean

much improved annual and lifetime uranium requirements. Con-

sidering the ease of which uranium content is recycled out of the

salt at the end of a batch cycle, this option is especially attrac-

tive.

Changing the 235U concentration in the LEU is also a possibledesire. The use of LEU with 20% 235 U attains the best uranium uti-

lization but this may entail difficult licensing issues for production

on a large scale. A level of 1015% might be a compromise that

would only slightly increase the required annual uranium. Oper-

ation on LEU alone is also possible and would require about 4.6%

LEU if the same starting fissile load of 3450 kg 235U is assumed.

Makeup fuel could be 56% LEU to still fall within current licens-

ing. Operation on LEU alone would require a shorter batch time of

perhaps 15 years or lower to limit the build-up of238 U and would

likely average a conversion ratio of about 0.7 instead of 0.8 and

require at least 1500 tonnes ore over two15 year cycles as opposed

to1000tonnes forthe standard30 year DMSR withuraniumrecycle

and 6400tonnes for a PWR once through. Such operation on LEU

alone or with only very small amounts of thorium brings aboutother potential advantages which will be examined in more detail

in Section7.6.


11/14



Another route for simple change is to lower the 30 year batch

time that the salt is used (22.5 years full power equivalent).As long

as uranium and hopefully transuranics are recycled into another

batch of salt, this greatly lowers the already impressive lifetime

uranium requirement. This would also lower the peak inventory

of fission products within the salt, one drawback to the DMSR

converter design. It should be mentioned that it is typically the fis-sion products with hours or days long half lives which are of most

concern in any accident scenario. These will be in similar concen-

trations for designs that remove fission products every few weeks,

every few years or even every few decades.

Another advantage of lowering the batch cycle time is an

improvement of quality of the uranium to be recycled. If a batch

time of 15 years is used, the recycled uranium is still very reac-

tive (8.3% 233U and 4.3% 235U) (Engel et al., 1980). It is also more

than needed to restart with clean carrier salt with a remainder

being available for several years worth of top up uranium addi-

tions. In ORNLs work with a 30 year cycle time, it was felt that

re-enrichmentof therecycleduraniumwouldbe necessarybecause

by the 30 year mark, the 233U + 235U fraction would be a bit low

(5.7% and 3.7% respectively). Two 15 year batch cycles with ura-nium recycle would likely drop the 30 year uranium utilization

down to 700 tonnes ore (600 tonnes U) or perhaps 625 t ore with

both uranium and transuranic recycle (1/10th that of LWR once

through).

With shorter batch cycle times, it is more likely that the carrier

salt itself would be recycled at a central facility by removing the

soluble fission products for longterm storage. The option also exists

to separate out the few long lived fission products such as 99Tc

for transmutation at the penalty of a slight increase in uranium

needs. Some of these, including 9 9Tc, do not form stable fluorides

so they cannot be simply placed back into the salt but they could

be embedded into future core graphiteelements for transmutation.

Thorium can also be recycled from the carrier salt but there is not

a strong economic incentive given its low cost and abundance. If it

is not recovered, then about 100 tonnes thorium per batch cycle is

needed (only about 3 million dollars worth).

Even with ORNLs 30yeardesign itis likelythat a plantrefurbish-

ment would be desired to extend plant life beyond 30 years. This

wouldentail replacingthe graphiteand perhaps the outervessel. Of

course, loweringpowerdensity to give 60 or more years of graphite

lifetime is possible but going to smaller cores, higher power densi-

ties andshortergraphitelifetimes might alsobe attractive to reduce

initial capital costs. Removing the need to replace graphite was

always one of the great simplifications of the DMSR designs but

one that is not tied to the many other proliferation and cost saving

advantages of the DMSR converter.

Higher power density will typically mean a lower salt volume,

thus lower fissile inventory and result in more neutron losses to

233Pa. However in the DMSR design these losses only amountedto 0.36% of neutron absorptions (0.18% to 23 3Pa which is doubled

to account for the neutron that created it from 232Th). For exam-

ple, with a smaller core and a total volume of salt of 25 m3 instead

of 100m3, losses to 233Pa would increase to 1.44% but this would

only drop the average conversion ratio by about 0.023 (from 0.8)

and only increase lifetime uranium needs by about 100tonnes ore.

Leakage of neutrons would be increased slightly in a smaller core

but there would also be a benefit of better 241Pu utilization since

in the DMSR study 4% of it decays before it is consumed. In terms

of losses to fission products, by also using 1/4 length batch cycle

of 7.5 years, the time averaged fission product concentration and

associated neutron losses would be approximately the same.

Such an operation of 1/4 volume, 1/4 batch time would also

mean justover1/4 the fissilestart-upinventory (somewhatmoreisneededto have a higherkinffora smaller diameter core).Thusabout

900kg 235U in LEU per GW(e) if other core parameters remained

the same. This would significantly lower capital costs and make

rapid deployment all the simpler (PWRs require about 5000kg per

GW(e)). While graphite replacement is not unduly complicated,

in subsequent sections an alternate method of easing graphite

replacement will be reviewed.

A final simple change to mention would be to switch from batch

replacementsof the salt to small but continuous salt replacements.If the fuel salt is replaced in small amounts equivalent to replacing

the entirevolumeover30 yearsthiswouldsaveon neutron lossesto

fissionproductsby about a 1.5ratio based on ORNL estimates. Thus

a 30 year continuous cycleis equivalent to a 20 year batchcycleand

lowers the needed uranium ore. This would only represent about

9 L of salt per day.

7.3. Simple tank of salt DMSR converter

One might consider attempting to remove graphite altogether,

simply running a large tank of salt as the core. The carrier salt itself

is a fair moderator of neutrons so a wide choice of neutron spec-

trum is still possible. There are several issues though, that decrease

the attractiveness of this simple core design. One is that

238

U res-onance absorptions will be of greater relevance, thus if using LEU

as starting fuel, it will be difficult to add very much thorium and

stillmaintain criticality. This means less contributionfrom the neu-

tron rich Th-233U chainand greater uranium oreneeds. As well, it is

unavoidablethatneutronlossestothecarriersaltwillbelargerthan

thecombinedsalt + graphitelosses of the traditional design. Studies

on early sphere within sphere graphite free designs (Alexander et

al., 1959) of the late 1950s fueled on93% 235U andthorium cangive

an indication of potential results. A long term average conversion

ratio of greater than 0.5 is unlikely with LEU + Th unless high fissile

concentrations and thus high specific inventories are employed.

High fissile starting loads could significantly affect start-up costs

and ability for rapid deployment.

A simple tank of salt also has no ability to have an under-

moderated outer blanket zone to help limit neutron leakage like

a graphite design can do. In fact, it is likely that any simple salt

tank design would require graphite liner/reflectors to shield the

outer vessel wall such as proposed in the MOSART (Ignatiev et

al., 2007) transuranic burner design. This graphite would likely

require replacement, which removes much of the simplification

advantage.

7.4. Pebble bed DMSR converter

Using graphite pebbles is a proposed change with great poten-

tial. Thepebbleswouldbe extremelysimple to manufactureas they

are only graphite and perhaps a pyrolytic coating. Simplification

would also extendto overall core designsince complicated topand

bottom plenum spaces and the need to lock together graphite ele-ments would not be required. Going to a higher power density and

a smaller core is also now straight forward as replacing graphite is

as simple as cycling outindividualpebbles or shutting down briefly

forbatch pebblereplacement. OakRidge alwaysconsidered theuse

of pebbles as a strongcandidatefor breeder designs andmanystud-

ies were performed on their use (Rosenthal et al., 1969, 1970). For

example in one study (Rosenthal et al., 1970) with the pure Th-233U cycle, a simple pebble bed design without any attempt at an

undermoderated outer zone and about the same fissile inventory

did almost as good for a breeding ratio as the standard MSBR (1.03

versus 1.06). Oak Ridge was under heavy pressure to have as high

a breeding ratio as possible, so pebble bed designs remained in the

background despite their many advantages.

Having an undermoderated outer zone to limit the small lossesof leakage neutrons is also still possible with pebble bed designs.

For example in ORNLs designs it was often proposed to have a


12/14



Fig. 6. An embodiment of a Pebble Bed DMSR Converter based on a modification of

design work from ORNL 4344.

graphite liner/reflector around the pebble bedto limit neutron flux

atthe outer vessel. This liner canbe simplymodifiedto bean under-

moderated neutron absorbing zone, for instance, having 50%of the

liner volume as salt channels.Fig. 6is a simple modification of an

ORNL engineering drawing (Rosenthal et al., 1969) to show this

proposed concept. Such an absorbing zone would mean smaller

cores with higher power densities are possible without significant

neutron leakage.

A random pebble bed does have a higher than ideal salt frac-

tion of 37% but the same neutron spectrum can be obtained simply

by having a lower fissile + fertile concentration in the salt to thus

give similar carbon to fissile ratios. There will undoubtedly be an

increase in the neutron losses to the salt but will have only a minor

effect on uranium utilization.

In general, the technique of using an undermoderated outer

region as proposed by ORNL works quite well in lowering neutron

leakage evenfor quitesmall diameter cores. By using thistechnique

it may prove optimal to switch any Single Fluid design (pebbles or

graphite logs) to an elongated cylinder geometry, without increas-

ing neutron leakage significantly. There are numerous advantages

to a longer, slender core. For one, top and bottom plenum spaceswherecore salts collect area challenge to designand take up a large

relative volume of salt. This would be minimized in a slender core.

Fabrication of a long slender vessel would also be simpler. Finally,

this might introduce new possibilities of in-situ decay heat man-

agement if no part of the core was too far from the outer vessel

wall. Decay heat could flow by conduction through the graphite

and salt in a manner similar to TRISO fueled PBMR or prismatic

core designs. Decay heat removal is typically done by the passive

drainingto dumptanks butthis mayoffer an interesting alternative.

7.5. Alternate carrier salt DMSR converter

Another simple concept but of perhaps great importance would

be to utilize an alternate carrier salt instead of the standard flibe(7LiF-BeF2). Flibe carrier salts are neutronically superior but the

enriched lithium and beryllium are expensive, beryllium is toxic

and they both lead to small but significant production of tritium

(roughly equivalent to that of CANDU Reactors). Managing tritium,

which can diffuse through hot metal walls of heat exchangers, was

always a significant part of ORNLs work on these systems. Their

choice of intermediate coolant salt (NaF-NaBF2) was made in part

to trap tritium and in general, while certainly not a show stopper,

tritium has always been a source of concern. It should be men-tioned that gas Brayton cycles are often now assumed the best fit

to molten salt reactors and these offer still further advantages for

tritium management because anytritium making it to the gas is far

easierto remove than from steam.Steamcycles however represent

an attractive off the shelf option, especially with the newest Ultra

SuperCritical cycles that could approach 50% efficiency if coupled

to molten salt reactors.

There are several carrier salt candidates that do not require

either 7Li or Be withan obvious choice being the very first salt used

in the Aircraft Reactor Experiment, NaF-ZrF4-(U,Th)F4. Otherpossi-

bilities include NaF-RbF-(Th,U)F4, RbF-ZrF4-(U,Th)F4NaF-(Th,U)F4although the later has a somewhat high melting point. There are

also many interesting choices that at least remove the need for

enriched lithium, the most likely choice being NaF-BeF2-(Th-U)F4.Therehas beenmuch recent examination of NaF-ZrF4 asa carrier

salt in transuranic waste burners. In those studies (Greenspan et

al., 2002) criticality was difficult to maintain due to the limited

solubility of PuF3. Carrier salts can typically carry 10 times more U

and/or Th so reaching criticality for a LEU converter design is not

a problem. There is also a higher solubility for PuF3 and any rare

earth trifluorides when UF4 or ThF4 is present (Ward et al., 1960).

This is important for converter designs in which trifluoride fission

products build up with time.

Invarious studies such as with NaF-ZrF4(Greenspan et al., 2002)

or LiF-NaF-BeF2(Ignatiev et al., 2007) or7LiF-NaF-(Th,U)F4(Nuttin,

2002) the increased neutron losses relative to flibe are not exces-

sive. Based on these studies, the non-tritium producing salts in a

graphite moderated designmight be expected to see 23 times the

losses of neutrons to the carrier salt as compared to flibe. In fact, if

the carrier salt has a significantly higher concentration of LEU + Th

than the 9.5% proposed for the DMSR, then a far lower salt volume

percentage is needed in core. In this case, some salt alternatives

could perhaps do as well or better than flibe. In ORNLs DMSR con-

verterdesign, neutron lossesto salt were only 1.53%so even iflosses

were a worst case of 5 times higher, the average conversion ratio

would drop from0.8 to about 0.67 and still require less than a third

the annual uranium ore needs of an LWR. At a natural uranium

cost of 100$/kg the annual LEU fuel costs ( Moir, 2007), including

enrichment, would only be 10.5 million (5 for ore, 5.5 enrichment)

per GW(e)year (0.29% tails, 110$/kgSWU) for this worst case. Even

natural uranium at 1000$/kggivesonly 56 million whichis roughly

current LWR fuelcosts including fabrication. Suchestimates of neu-

tron losses and associated uranium utilization are obviously onlycrude estimates.

7.6. An LEU only DMSR converter

As briefly discussed in Section 7.2, operationon LEUalonewith-

outthoriumis an interestingpossibleoption. A disadvantageis that

it removes thefractionof fissions comingfrom the neutron rich Th-233U cycle. In the traditional DMSR core and looking at the 15 year

mid point, the weighted average of (neutrons per fissile absorp-

tion) is 1.95 for 235U + 239Pu + 241Pu while 233U alone is 2.24. This

gives an overall average of 2.10 (weighted by fission fraction). If

operation would be on LEU alone then the average should drop

down to about 1.95 which suggests a roughly 0.15 drop in con-

version ratio (0.80.65). However both options would start withroughly the same conversion ratio of 0.8 (both start with 235U as

the only fissile with an of 1.99). For the LEU + Th case, the conver-


13/14



sion ratio rises over the first several years due to 233 U production,

then drops as fissionproducts build up.For LEUalone, it would start

around 0.8 and drop slowly as fission products build up. The long

term average conversion ratio for LEU only should thus be close to

0.7 whereas with shorter batch cycles this could be closer to 0.75.

This means the penalty to uranium utilization is smaller than one

might at first suspect.Running on LEU alone opens up some unique advantages that

may not have been examined in the past. First, while UF4and ThF4tend to behave similarly in carrier salt eutectics, salt mixes with

only UF4 typically have lower melting points or conversely, allow

higher UF4 concentrations and still remain below the traditional

525 C melting point limit for use with Hastelloy N. Second, with-

out thorium (or only havingsmall amounts) means the core design

is no longer constrained regardinglimiting neutron losses to 233Pa.

This means that thepowerdensity canbe increasedgreatly (lower-

ingthe start-up fissileinventory) andthere is no longera constraint

to have a more thermalized neutron spectrum (as previously

reviewed, the DMSR is a softer spectrum than the MSBR to lower

neutron losses to Pa). It should be mentioned though that there are

still advantages to a quite well thermalized spectrum though sincefor 239Pu and 235 U are somewhat better. As well, a soft spectrum

means the absorption and fission cross sections for all Pu isotopes

are larger which means it will stay at low concentrations which

may prove important given the limited solubility of PuF3.

Finally, in terms of proliferation resistance this design might

represent the highest obtainable in any reactor design. There is

no 233Pa present that could somehow be chemically separated to

yield 233U. While plutonium is present and will need to be safe-

guarded, it will be of extremely low quality in terms of weapons

used (much poorer than LWR spent fuel, etc.). The only window of

opportunity for misuse would be in the first year or two of opera-

tion when the isotopic quality of the contained plutonium is more

of a concern. This is easilycountered however by simplystartingall

reactors using a mix of LEU and solid reactor spent fuel plutonium

such that the plutoniumpresent is always of very poor quality (and

as always, very difficult to remove).

7.7. DMSR converter design summary

In general for DMSRconverter designs,the potentialdoes appear

obvious forthis underappreciatedclassof moltensalt reactors. The

pure Th-233U cycle does lead to the ultimate in resource utilization

but if simple converter designs can perform just as well in other

categories and excel in overall cost, minimized R&D and prolifera-

tion resistance, it is only right they receive more attention. These

simplified designs can also be looked upon as merely a 1st gen-

eration that can get up and running quickly with minimized R&D

costs. Fuel processing capabilities or simple core changes can be

made later to change to consuming thorium only. Again this hingeson the arguments around Peak Uranium but as shown inTable 2,

even with a natural uranium price of $5000/kg, operations would

still be economical and open up virtually inexhaustible supplies,

fromnumerousunconventional sources suchas simple granite rock

or sea water.

8. Summary

A goal of this report has been to show the great versatility and

potentialof moltensaltdesigns.This may have come at the expense

of appearing unfocused. With that in mind it would appear wise to

put forth three designs considered to be leading candidates and to

act as starting points for further modeling. A starting core power of

200MW(e) appears an appropriate compromise between economyof scale and needs of component development costs (pumps, IHX,

turbines, etc.).

Case I would bea Th-233U cycle, graphitefree, break even design

of the simple tube withintube variation(LeBlanc, 2008) employing

a metal barrier between core and blanket salts of either recently

developed Hastelloy alloys or a high molybdenum alloy. The later

being more difficult to fabricate but likely having a much longer

in core service life. To break even on breeding, a fissile concen-

tration of about 0.3% 233UF4 will likely be required which impliesa tube diameter of about 85 cm and core length of 4 meters. At a

200 MW/m3 power density this gives 450MW(th) and 200 MW(e).

The surrounding blanket salt would be 73%7LiF27%ThF4 of about

1 m thickness and contained in an outer Hastelloy N vessel. Start-

up fissile needs would be about 800 kg/GW(e) or 160 kg for one

200 MW(e) unit. Alternative start-up would be on LWR transuran-

ics or by LEU as previously reviewed. Fission product processing

of the fissile salt being by vacuum distillation on a 6 month

cycle. The fertile salt would only receive periodic fluorinations to

transfer produced 233U to the central fuel salt. At a small added

expense, further processing to remove and recycle the minor

amounts of Pu, Amand Cm fromthe fissionproducts is likelydesir-

able.

Case 2 would be a LEU + thorium converter reactor with onlyminor changes from Oak Ridges design. The proposed changes

that would appear to entail little added R&D would be a mod-

estlyhigher(2) power density andemploying thealternate carrier

salt 46%NaF33%RbF21%(U+ Th)F4 with a melting point under

500 C. Core size for 200 MW(e) and the modestly higher power

density would be under 4 m width and height. This alternate car-

rier salt is inexpensive and produces no tritium but would require

confirmation of adequate trifluoride solubility and compatibil-

ity with Hastelloy N (both likely). It should be mentioned that

rubidium currently has almost no annual production but can be

found as a by product of other alkaline metal mining (Williams,

2006).

A doubling of power density would mean graphite replace-

ment every 1215 years (11.25 full power years) but would

substantially reduce the size and cost of the core along with

the starting fissile inventory. Salt replacement at the same time

appears wise with the removal and reuse of all uranium and

likely all transuranics (Np, Pu, Am, Cm) into the subsequent batch.

This salt choice has a much higher fissile concentration than in

the DMSR study and thus a lower percentage of salt relative to

graphite is employed. Thus it should do about as well on uranium

utilization as the neutronically superior, but more voluminous,7LiF-BeF2 carrier salt. Cost and rapid production potential of the

NaF-RbF carrier salt would also see a large advantage over 7LiF-

BeF2.

Case 3 would be a pebble bed design employing an undermod-

erated outer zone to limit neutron leakage as discussed in Section

7.4. Fissile concentration in the carrier salt would be lowered by

a factor of 3 to counter the tripling of salt fraction in a randompebble bed compared to the DMSR (0.37 vs. 0.13). This should

result in a similar neutron spectrum. Again an alternate carrier

salt is possibly an improvement and since the fissile/fertile con-

centration is low, more carrier salt choices are available, such as

NaF-ZrF4, RbF-ZrF4 or NaF-BeF2. As well, trifluoride solubility is

of far lesser concern as fission products and PuF3 remain more

diluted.

A 200 MW(e) version with three times the average power den-

sity of the DMSR would be a pebble bed core 3.9 m in diameter

surrounded by an outer undermoderated zone of 0.5 m of solid

graphite with 50% salt channels. Total salt volume would be 26 m3

in the reactor with perhaps 4 m3 out of core. This would result in

a starting fissile inventory of about 350 kg 235 U (as LEU). A shorter

batch cycle time of about 57.5 full power years would give a sim-ilar fission product to fissile ratio profile as the DMSR. Uranium

utilization will be slightly higher than the DMSR using flibe and


14/14



modestly higher still with an alternate carrier salt but still only a

small fraction of LWR use.

For comparison, a 1000 MW(e) version and the same 3 times

average power density would be a pebble bed core of 6.6 m diam-

eter or 7.6 m including the undermoderated zone. This would only

require approximately 1100 kg of235 U for start-up (less than one-

third than for the DMSR and one-fifth that of a LWR).

9. Conclusions

As is hopefully now evident, molten salt reactor designs offer

great flexibility and advantages for any priorities one may set for

advanced reactors. Overall power costs of the traditional Single

Fluid Breeder design have been estimated to be lower by a small

amount than LWR costs (Rosenthal et al., 1972; Moir, 2002; Moir

and Teller, 2005), such that the great simplifications in design and

fuel processing proposed here should lead to a clear advantage.

Overall safety sees a multitude of advantages over other reactor

designs and denatured operation or even LEU only designs can

be employed if perceived proliferation issues of the pure Th-233U

cycle become an issue. Design and modeling work is ongoingby the author and others on these designs and numerous others

giving the versatility to adapt to unforeseen design obstacles.

While at present, government and industry support is sorely

lacking worldwide, the research and development needs (Forsberg,

2006b) are far less than many may imagine. Perhaps ORNLs

greatest legacy in this respect has been their dedication to doc-

ument all aspects of their work. This great wealth of technical

information is now accessible at www.ornl.gov/info/library, or

www.energyfromthorium.com and is being actively reexamined.

While the lack of after sale profits in terms of solid fuel fabrication

may require a different business model to attract corporate inter-

est, the potential rewards are indeed great for any government,

corporation or agency willing to take a leading role in this vital

effort.

References

Alexander,L.G., et al.,1959. Nuclear characteristicsof spherical,homogeneous,two-region, molten-fluoride-salt reactors. ORNL 2751.

Bowman, C.D., August 1998. Once through thermal spectrum accelerator drivensystem for LWR waste destruction without reprocessing. Report ADNA 98-04.

Cury, R., 2007. Etude metallurgique des allianges Ni-W et Ni-W-Cr: relation entreordrea courte distance et durcissement. PhD Thesis UniversiteParis XII.France.

DiStefano, J.R., et al., March 1995. Materials considerations for molten saltaccelerator-based plutonium conversion systems. ORNL/TM-12925/R1.

Engel, J.R.,et al.,August 1978.Molten saltreactorfor efficientnuclear fuelutilizationwithout plutonium separation. ORNL-TM-6413.

Engel, J.R., et al., July 1980. Conceptual design of a denatured molten salt reactorwith once-through fueling. ORNL-TM-7207.

Forsberg, C.W., 2006a. Developments in molten salt and liquid salt cooled reactors.Proceedings of the ICAPP.

Forsberg, C.W., 2006b. Molten Salt Reactor Technology Gaps Proceedings of ICAPP.Furukawa, K., et al., 1990. Summary report: thorium molten-salt nuclear energysynergetics. J. Nucl. Sci. Technol. 27, 11551178.

Furukawa, K., 1992. The Combined System of Accelerator Molten-Salt Breeder(AMSB) and Molten-Salt Converter Reactor (MSCR). Japan-US Seminar on ThFuel Reactors, Nara.

Furukawa, K., et al., 2008. A road map for the realization of global-scale thoriumbreeding fuel cycle by single molten-fluoride flow. Energy Conv. Manag. 49,18321848.

Gat, U.,et al., 1992. MoltenSaltReactors forBurningDismantledWeaponsFuel. Nuc.Tech. 100 (3), 390394.

Gat, U., et al., 1997. Molten Salt Reactorssafety options galore. In: Proc. of anInternational ANS Topical Meeting on Advanced Reactor Safety, Orlando FL.

Greenspan, E., et al., 2002. Optimization of a Molten Salt Transmuter Reactor.PHYSOR, Seoul, Korea.

Greenspan, E., et al., 2007. Transmutation performance of Molten Salt versus solidfuel reactors. In: 15th Int. Conf on Nuclear Engineering, Nagoya Japan ICONE15.

Ignatiev,V., etal., 2007. Progressin Developmentof Li,Be, Na/F MoltenSaltActinideRecycler and Transmuter Concept. Proceedings of the ICAPP 2007.

Keiser, J.R., et al., April 1977. The Corrosion of Type 316 Stainless Steel to Li 2BeF4.ORNL-TM-5782.

Koger, J.W.,December 1972.Alloy Compatibility withLiF-BeF2SaltsContaining ThF4and UF4 . ORNL-TM-4286.

LeBlanc, D., May 2008. Molten Salt Nuclear Reactor. U.S. Patent Application, Ser12/118,118.

Lecarpentier, D., 2001. Le Concept AMSTER, Aspects Physiques et Surete. Ministerede lEducationNationalConservatoireNationaldes Artset metiers. Ph.DThesis.

MacPherson, H.G.,et al., 1958.Molten SaltProgram Quarterly Progress Report ORNL2551.

McNeese, L.E., February 1971. Engineering Development Studies for Molten SaltBreeder Reactor Processing, No. 2. ORNL-TM-3137.

Merle-Lucotte,E., et al.,2007a. Optimized transition fromthe reactorsof second andthird generation to the Thorium Molten Salt Reactor. Proceedings of the ICAPP2007.

Merle-Lucotte,E., 2007b.Introductionto thephysicsof molten saltreactorsin Mate-rials Issues for Generation IV Systems. Status, open questions and challenges.

NATO Advanced StudyInstitute on Materials forGeneration IV Nuclear Reactors(MATGEN-IV), France.

Merle-Lucotte, E., 2008a. Optimization and simplification of the concept of non-moderated Thorium Molten Salt Rector. In: International Conference on thePhysics of Reactors PHYSOR.

Merle-Lucotte, E., 2008b. Le Cycle Thorium en Reacteurs A Sels Fondus Peut-ilEtre Une Solution au Problem Energetique du XXIeme Siecle? Le Concept deTMSR-NM. dHabilitation a Dirirger les Rechers de LInstitute Polytechnique deGrenoble.

Mourogov, A., Bokov, P.M., 2005. Potentialities of the Fast Spectrum Molten SaltReactor Concept: REBUS-3700. ICENES2005, Brussels.

Moir, R.W., 2002. Cost of Electricity from Molten Salt Reactors. Nucl. Technol. 138,93.

Moir, R.W., Teller, E., 2005. Thorium fueled underground power plant based onmolten salt technology. Nucl. Technol. 151, 334.

Moir, R.W., 2007. Recommendations for a Restart of Molten Salt Development.ICENES2007, Istanbul.

Nuttin, A., 2002. Potentialits du concept de racteur sels fondues pour une

production durable denergie nucleaire basee sur le cycle thorium en spectreepithermique. Ph.D Thesis. LUniversite Joseph Fourier-Gernoble I.Peterson, P.,et al.,2008. Design, Analysis andDevelopment of theModular PB-AHTR.

ICAPP 08, Anaheim CA.Robertson, R.C., et al., August 1970. Two-Fluid Molten Salt Breeder Reactor Design

Study (Status as of Jan 1, 1968). ORNL 4528.Rosenthal, M.W., et al., February 1969. Molten Salt Reactor Program Semiannual

Progress Report. ORNL-4344.Rosenthal, M.W., et al., August 1970. Molten Salt Reactor Program Semiannual

Progress Report. ORNL-4548.Rosenthal,M.W., et al.,August 1972.Molten SaltReactor Program The Development

Status of Molten Salt Breeder Reactors. ORNL-4812.Scott, C.D.,et al.,January 1966.Preliminary design studyof continuousfluorination-

vacuum-distillation system forregenerating fuel and fertile streams in a moltensalt breeder reactor. ORNL 3791.

Shaw, M., September 1972. U.S. Atomic Energy Commission An Evaluation of theMolten Salt Breeder Reactor. U.S. Atomic Energy Commission, WASH 1222.

Sorensen, K., 2007. Private Communication.Ward, W.T., et al., 1960. Rare earth and yttrium fluorides: solubility relations in

various molten NaF-ZrF4 and NaF-ZrF4-UF4 solvents. J. Chem. Eng. Data 5 (2),137142.

Weinberg,A.M.,et al., 1970. Entirevolume onmolten salt breeder reactor.Nuc. App.Tech. 8 (February).

Williams, D.F., March 2006. Assessment of Candidate Molten Salt Coolants for theAdvanced High Temperature Reactor (AHTR). ORNL/TM-2006/12.

Zinkle, S.J., Ghoniem, N.M., 2000. Operating windows for fusion reactor structuralmaterials. Fusion Eng. Des. 5152, 5571.

molten salt leblanc

Documents